Interactive Rule Mining in Knowledge Bases
Rule Mining is the process of automatically discovering logical rules in knowledge bases (KBs). This is a challenging endeavor, for two reasons: First, KBs usually provide no negative information, meaning that the mining process cannot use counter-examples for rules. Second, today’s KBscontain millions of facts, so that the search space for the mining is of considerable size. State-of-the-art systems cut down the search space by systematically pruning unpromising rule refinements. We propose to make this process transparent to the user through an interactive demo.